Classification of Cataract Disease with a DenseNet201 Based Deep Learning Model
Halıt Çetıner, İbrahim ÇETİNER
Abstract
Cataracts are among the most serious eye diseases and can cause blindness if left untreated. Since it is a treatable disease, professional knowledge of specialist ophthalmologists is needed. Ophthalmologists need to analyze images of the eye to detect clinical cataracts in an early stage. Detection of cataracts at an early stage prevents the disease from progressing and causing serious costs such as blindness. At this point, it is a tiring and costly process for specialist ophthalmologists to constantly check their patients. It is not possible for ophthalmologists to constantly monitor their patients. Due to the stated problems, in this article, a study was carried out to develop a deep learning model that helps specialist ophthalmologists through cataract images. In the developed model, an automatic classification of images with normal and cataract lesions was performed by proposing a model based on pre-trained neural networks. During the development of the proposed model, the performance of the classification process was increased by making fine adjustments to the pre-trained neural network called DenseNet201. To compare the performance level of the proposed model, the results obtained from the model consisting of the basic DenseNet201 structure without using any additional layers were used. When both models are evaluated, it has been shown that the proposed deep learning model achieves 10% more success than the basic DenseNet201 deep learning model. The proposed model can be used as an auxiliary tool for doctors in different health problems such as cataracts, which are commonly encountered today.